Pentagon shock as AI system breaks into secret government networks during test, official confirms (opens original article in a new tab)

An AI model developed by Anthropic identified vulnerabilities in U.S. government systems during a test, prompting security concerns and government restrictions on the technology.

• An AI model from Anthropic found vulnerabilities in U.S. government systems during a test
• The test was part of Project Glasswing aimed at securing global software
• The U.S. government restricted access to Anthropic's AI models over security concerns
• Cybersecurity experts criticized the government's directive as potentially harmful